WO2023140767A1

WO2023140767A1 - Beam scanning with artificial intelligence (ai) based compressed sensing

Info

Publication number: WO2023140767A1
Application number: PCT/SE2023/050037
Authority: WO
Inventors: Tamas Borsos; András RÁCZ; Máté Szebenyei
Original assignee: Telefonaktiebolaget Lm Ericsson (Publ)
Priority date: 2022-01-19
Filing date: 2023-01-16
Publication date: 2023-07-27

Abstract

A method performed by a user equipment, UE, (1600) in a network includes obtaining (501) a power allocation matrix derived based on a machine learning process, wherein each row of the power allocation matrix defines a power allocation vector specifying how much power is allocated to each beam of a number of access points, APs, each AP having a number of beam directions per AP. The method further includes measuring (503) combined reference signals received by the UE for each power allocation combination transmitted according to the power allocation matrix. The method further includes calculating (505) a sparse vector D using a compressed sensing algorithm based on a number of the combined reference signals measured. The method further includes determining (507) best beam directions based on the sparse vector D. The method further includes signaling (509) the best beam directions towards a network node in the network.

Description

Beam Scanning With Artificial Intelligence (Al) Based Compressed Sensing

[0001] The project leading to this application has received funding from the European Union’s Horizon 2020 research and innovation program under grant agreement No 101015956.

TECHNICAL FIELD

[0002] The present disclosure relates generally to communications, and more particularly to communication methods and related devices and nodes supporting wireless communications.

BACKGROUND

[0003] In high frequency bands, radio signal propagation is suffering from large pathloss per distance and signals can get easily blocked by objects in the physical space, which altogether make the radio communication in these high frequency bands much more volatile and difficult as compared to lower bands. This is illustrated in Figure 1 where the UE does not receive signals from all access points/base stations due to blockages and/or receives partially blocked signals. One solution to combat these problems is to direct the radio signal via beamforming in the direction where line of sight (LOS) or close to LOS propagation is possible.

[0004] Finding the best beam in all situations is a cornerstone of the beamforming problem, which can be divided into two main phases: (1) initial beam search and (2) beam tracking/update. At initial beam search when the UE has no connectivity to the network, it has to scan all possible beam directions to find the best one. For this purpose, the gNodeB transmits so called SSB (Synchronization Signal Blocks) signals, and the UE is doing a sweeping through the possible beams, measuring them all one-by-one. The UE scans for SSB signals, measures and selects the strongest beam direction. This can be a lengthy procedure and requires much transmit resources and time. If beamforming is available at the UE as well, this can multiply the possible beam combinations and associated scan time. The problem can become even more severe in Distributed massive MIMO (D-MIMO) settings where there are a large number of access points (APs) (typically more than UEs) each with a couple of transmit antennas and possible beam directions.

[0005] In the beam tracking problem, the best beam direction may need to be updated as it may change in time, e.g., as the user is moving or other users or objects are moving in the surrounding.

[0006] There currently exist certain challenge(s). Existing beam sweeping procedures are quite resource consuming both in terms of the transmission of reference signals (e.g., SSB signals) and in terms of scanning time and resources at the UE. These procedures can introduce significant delay e.g., during the initial access procedure.

[0007] The beam sweeping problem becomes even more severe in Distributed massive MIMO systems, where there are a large number of distributed antennas with many potential beam combinations. Finding the best beam with a full scan in such a setting would be prohibitively complex and time consuming.

SUMMARY

[0008] Various embodiments use compressed sensing for beam sweeping such that the best beam directions from all possible APs can be determined from a few transmissions without requiring a full scan search. The APs can transmit the same reference signal in all directions using the same time frequency resource and varying the transmit power allocations in a few combinations. The UE needs to measure only the composite received signal from all directions, it does not have to identify beams individually. The best beam direction(s) and corresponding channel gains can be determined using compressed sensing (CS) computation done either at the UE or at the network.

[0009] The required transmission combinations are optimized to the particular environment by using machine learning. Thereby the required number of measurements can be significantly smaller as compared to applying the conventional compressed sensing algorithms.

[0010] According to some embodiments, a method performed by a user equipment, UE, in a network includes obtaining a power allocation matrix derived based on a machine learning process, wherein each row of the power allocation matrix defines a power allocation vector specifying how much power is allocated to each beam of a number of access points, APs, each AP having a number of beam directions per AP. The method includes measuring combined reference signals received by the UE for each power allocation combination transmitted according to the power allocation matrix. The method includes calculating a sparse vector D using a compressed sensing algorithm based on a number of the combined reference signals measured. The method includes determining best beam directions based on the sparse vector. The method includes signaling the best beam directions towards a network node in the network.

[0011] Certain embodiments may provide one or more of the following technical advantage(s). Faster beam scan times both at initial access and at beam tracking may be achieved. Thus, there may be no need to do full scan search. There may be no need to use dedicated reference signals (e.g., SSBs) for each AP and each beam direction. Thereby less radio resources are spent on beam reference signal transmissions. This leads to less time and energy being spent at the UE for scanning.

[0012] According to some other embodiments, a method performed by a network node in a network includes transmitting a plurality of reference signals on beams on a plurality of same time-frequency resources in all directions in accordance with power allocation vectors of a power allocation matrix. The method includes receiving, from a UE, composite signal strengths received for each power allocation vector transmitted. The method includes calculating a sparse vector D using a compressed sensing algorithm based on a number of the composite signal strengths measured. The method includes determining best beams to use to transmit to the UE based on the sparse vector D.

BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The accompanying drawings, which are included to provide a further understanding of the disclosure and are incorporated in and constitute a part of this application, illustrate certain non-limiting embodiments of inventive concepts. In the drawings:

[0014] Figure 1 is an illustration of beam scanning and tracking where some beams are blocked by objects;

[0015] Figure 2 is an illustration of compressed sensing expressed as an undetermined set of linear equations according to some embodiments;

[0016] Figure 3 is an illustration of applying compressed sensing (CS) to beam scanning according to some embodiments;

[0017] Figure 4 is an illustration of transmission of beam scanning reference signals according to some embodiments;

[0018] Figure 5 is a flow chart illustrating operations of a UE according to some embodiments of inventive concepts;

[0019] Figure 6 is a flow chart illustrating operations of a network node according to some embodiments of inventive concepts;

[0020] Figure 7 is an illustration of best beam indexing during test measurements along different routes according to some embodiments;

[0021] Figure 8 an illustration of a number of beams with a signal strength > -100 dBm;

[0022] Figure 9 is an illustration of a power allocation dictionary according to some embodiments;

[0023] Figures 10A and 10B are illustrations of estimated best beam directions with a number of measurement combinations set to 25 according to some embodiments; [0024] Figures 11 A and 1 IB are illustrations of estimated best beam directions with a number of measurement combinations set to 30 according to some embodiments;

[0025] Figure 12 is a block diagram illustrating an autoencoder architecture for learning power allocation dictionary for a specific environment according to some embodiments;

[0026] Figure 13 is an illustration of estimated best beam directions with autoencoder learning with a number of measurement combinations set to 10 according to some embodiments;

[0027] Figures 14A and 14B are illustrations of estimated best beam directions with principal component analysis (PCA) decomposition with a number of measurement combinations set to 15 according to some embodiments;

[0028] Figure 15 is a block diagram of a communication system in accordance with some embodiments;

[0029] Figure 16 is a block diagram of a user equipment in accordance with some embodiments

[0030] Figure 17 is a block diagram of a network node in accordance with some embodiments;

[0031] Figure 18 is a block diagram of a host computer communicating with a user equipment in accordance with some embodiments;

[0032] Figure 19 is a block diagram of a virtualization environment in accordance with some embodiments; and

[0033] Figure 20 is a block diagram of a host computer communicating via a base station with a user equipment over a partially wireless connection in accordance with some embodiments in accordance with some embodiments.

DETAILED DESCRIPTION

[0034] Some of the embodiments contemplated herein will now be described more fully with reference to the accompanying drawings. Embodiments are provided by way of example to convey the scope of the subject matter to those skilled in the art, in which examples of embodiments of inventive concepts are shown. Inventive concepts may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of present inventive concepts to those skilled in the art. It should also be noted that these embodiments are not mutually exclusive. Components from one embodiment may be tacitly assumed to be present/used in another embodiment.

[0035] Independent of beamforming, the compressed sensing or compressive sensing (CS) theory is a relatively recently explored scheme, which has found its application in different areas more recently. For example, it has been successfully applied in imaging applications (e.g., in medical applications of magnetic resonance imaging (MRI)) or in radar signal processing.

[0036] Compressed sensing is a signal processing technique which says that if a signal is sparse in some domain, then the signal can be fully reconstructed from a fewer number of measurement samples than what would be required by sampling theory. Being sparse means that the signal contains a lot of zero elements when it is expressed in a certain domain or basis. For example, a signal can be sparse in the Fourier domain containing only a few non-zero Fourier components or an image can be sparse when transformed into the Discrete Fourier or Discrete Cosine Transform domain.

[0037] Explaining CS in an algebraic way, the CS problem can be formulated as solving an under determined set of linear equations. The setup is illustrated in Figure 2. Vector X is the encoded sparse signal with dimension of N and is a representation of the signal sample, vector Y is the measurement vector (i.e., original signal samples) with dimension M, where M« N and B is the measurement matrix (e.g., referred to as a basis matrix or dictionary) having size M x N. In principle, any basis is ok (e.g., does not need to be orthogonal). Vector X has K number of non-zero elements, but the position of zeros are unknown. What is known is vector Y, i.e., the measurement samples, and the measurement matrix B and the signal X is of interest. Since the number of equations (M) is less than the number of unknowns (N), the solution cannot be found with the classical linear equation solving.

[0038] Compressed sensing provides algorithms which can still solve these equations under the assumption that there are many zero elements in X. These algorithms attempt to find the solution with the constraint that the L0 or LI norm of vector X is minimized.

[0039] Previous work investigated low-overhead beam scanning solutions particularly, where the beam scanning method is enhanced with some side-information to avoid full scan default search. One of the approaches investigated was the application of CS theory for beam scanning. In one solution different linear combinations of GOB (Grid of Beams) beams are used to form composite beam patterns, where the number of different combinations is smaller than the number of all possible beam directions. The composite beam patterns are achieved by applying appropriate beamforming codebooks. In another variant, the beam pattern is kept fixed and only the beam energy is varied in a few combinations. The benefit of this approach is that phase coherence is not required.

[0040] One difference between the previous work and the various embodiments described herein is that the base CS method is extended with Artificial Intelligence (Al) to enable the learning of the optimal encoding dictionary (i.e., the beam forming codebook or beam power allocation dictionary) based on measured data. This enables the CS encoding to be adapted to the particular environment and thereby achieve better compression levels (i.e., less beam scanning measurements required) as compared to the base CS method. Furthermore, the previous work assumed a single point MIMO scenario and cannot consider and optimize for a massive distributed MIMO (D-MIMO) scenario, whereas the various embodiments of the subject matter described herein can consider and optimize for massive D-MIMO.

[0041] The various embodiments of the subject matter described herein use compressed sensing for beam sweeping such that the best beam directions from all possible APs can be determined from a few transmissions without requiring a full scan search. The APs can transmit the same reference signal in all directions using the same time frequency resource and vary the transmit power allocations in a few combinations. The UE needs to measure only the composite received signal from all directions, it does not have to identify beams individually. The best beam directi on(s) and corresponding channel gains can be determined using CS computation done either at the UE or at the network.

[0042] The required transmission combinations are optimized to the particular environment by using machine learning. Thereby the required number of measurements can be significantly smaller as compared to applying the conventional compressed sensing algorithms.

[0043] Application of CS theory basis for beam scanning.

[0044] Applying the CS theory on the beam scanning problem in the various embodiments is illustrated in Figure 3. The underlying observation and assumption is that the signal propagation is sparse in the spatial domain (in high mmwave, high frequency bands). This means that the vector that includes the channel gains from all APs and all directions would be sparse (i.e., it will have non-zero elements only in a few dimensions). As a result, the vector is a sparse vector where only a few directions are non-zero. In Figure 3, the vector with the channel gains is denoted as vector D, which has dimensions of N=L*B, where L is the number of APs and B is the number of beam directions per AP. Such a D_L vector exists for all UEs, hence the indexing i. In other words, the vector D includes all APs and all directions.

[0045] The measurement matrix P is a power allocation matrix of dimension M x N in this case, which means that each row of the matrix describes a power allocation vector, which specifies how much power is allocated to each beam for the transmission of the reference signal in the system. Each row specifies a different power allocation, which can be executed one-by- one as a transmission. The matrix multiplication of P and D is executed by the transmission of the beam scan reference signal from all APs in all directions with a power allocation specified by the given row of P. Note that all beams would transmit the same signal form and the multiplication operation would essentially happen over the air via the physical mixing of the signals. The combined received signal at the UE will be the corresponding element in vector Yi. There will be as many transmissions as the number of rows (i.e., the number of different power allocation combinations).

[0046] In order the summation of the signals over the radio interface to happen, it is necessary to transmit synchronously on the different beams. The time synchronization does not have to be phase coherent, a symbol level synchronization would be sufficient. In case of a D- MIMO system such a time synchronization is feasible to assume and would be needed for the coordinated transmissions anyway.

[0047] Then the vector D_L can be inferred from the observed measurements Yi having a dimension M x 1 of original signal samples. This calculation can be performed by using one of the reconstruction CS algorithms, such as Orthogonal Matching Pursue (OMP) or Iterative Hard Thresholding (THT), Basis Pursuit (BP), etc. The goal is to reconstruct the sparse signal from as few measurement combinations as possible (i.e., as few rows in matrix P as possible).

[0048] The power allocation matrix P is determined by the network and would be the same for all UEs. In other embodiments, the matrix P can be obtained by a learning process where the most “optimal” power allocation vector is determined by fitting to the actual measurement data. In CS theory there are algorithms that can be used for such learning, also called dictionary learning in CS.

[0049] For dictionary learning of matrix P, learning measurement samples are collected where the sparse D vectors are directly measured, which means to execute the classical beam scanning procedure. Based on the collected samples, the learning algorithm finds the optimal power allocation matrix such that the D vectors can be reconstructed from as few measurements as possible (i.e., from the shortest possible measurement vector Y). Via the learning the algorithm would try to find the most “optimal” P matrix, which allows the easiest reconstruction for the most typical beam combinations.

[0050] System realization of CS based beam scanning

[0051] The transmission of beam reference signals in the various embodiments is illustrated in Figure 4. Each AP transmits the same reference signal on the same time-frequency resource in all directions (as illustrated with the same pattern resource grids 400 in Figure 4). What differs between the different beam transmissions is the allocated power per beam 402 (indicated with the different size beams in the figure). [0052] The different power allocation combinations are according to the power matrix P in the CS equation. The different power allocations (i.e., the different rows of matrix P) are rotated continuously (going row-by-row of the matrix P) and the reference symbols are transmitted according to these power allocations periodically.

[0053] The UEs measure the received combined signals by measuring the power level of the signal for each power allocation combination one-by-one. It is possible to do some time averaging between multiple transmissions of the same power combination in order to reduce the effect of fast fading, for instance.

[0054] The transmission periodicity of the different power combinations is determined a- priori and can be signaled to UEs on the system information.

[0055] There can be two main variants of the solution depending on where the CS computation is performed at the UE or at the network side.

[0056] UE side CS computation

[0057] In this case, the power allocation matrix P needs to be signaled to the UE and the UE calculates the sparse vector D to determine the feasible beam directions and it signals back to the network the best beam directions. (It only needs to signal back the vector indexes, i.e., beam indexes that are above a certain threshold instead of the complete D vector.)

[0058] Figure 5 illustrates operations the UE performs in UE side CS computation. The UE 1600 (implemented using the structure of the block diagram of Figure 16) will be used in describing the flow chart of Figure 5. For example, modules may be stored in memory 1610 of Figure 16, and these modules may provide instructions so that when the instructions of a module are executed by respective UE processing circuitry 1602, processing circuitry 1602 directs the UE to perform respective operations of the flow chart.

[0059] Turning to Figure 5, in block 501, the UE 1600 obtains a power allocation matrix derived based on a machine learning process, wherein each row of the power allocation matrix defines a power allocation vector specifying how much power is allocated to each beam of a number of access points, APs, each AP having a number of beam directions per AP.

[0060] In block 503, the UE 1600 measures combined reference signals received by the UE for each power allocation combination transmitted according to the power allocation matrix. In some embodiments, the UE 1600 measures the combined references signals by measuring a received composite signal strength for each power allocation combination. The UE 1600 may transmit the measurement results to a network node in some embodiments.

[0061] In block 505, the UE 1600 calculates a sparse vector D using a compressed sensing algorithm based on a number of the combined reference signals measured. The calculation has been described above and experimental results are provided hereinbelow. In some embodiments, the UE 1600 calculates the sparse vector D based on a machine learning process. For example, the UE 1600 may calculate the sparse vector D using one of an orthogonal matching pursue, OMP, an iterative hard thresholding, IHT, and a basis pursuit, BP. In some of these embodiments, the UE 1600 calculates the sparse vector D based on dictionary learning.

[0062] In block 507, the UE 1600 determines best beam directions based on the sparse vector D. In some embodiments, the UE 1600 determines the best beam directions by determining which beam directions are above a threshold and indicating that the beam directions that are above the threshold are the best beam directions.

[0063] In block 509, the UE 1600 signals the best beam directions towards a network node in the network.

[0064] The network knows the mapping between the vector index and the actual physical beams, so it can map the index to the physical beams which should be used to transmit toward that UE.

[0065] Network side CS computation

[0066] In these embodiments, the power allocation matrix P does not need to be signaled to the UE, the UE only needs to know the index of the power allocation combination that is currently being used for the reference signal transmission and measure on the composite received signal (similarly as it was in the previous case). The repetition pattern and periodicity of the different power allocation combinations can be distributed to the UE once, either via dedicated signaling or via system information broadcast.

[0067] The UE measures the received composite signal strength (optionally performing some time averaging) and reports the result back to the network (i.e., it reports the measurement vector Yi). The network performs the CS algorithms to reconstruct the sparse matrix D and identifies the best beams to be used to transmit to the given UE. This same calculation is performed for each UE individually.

[0068] Figure 6 illustrates operations the network node performs in network side CS computation. The network node 1700 (implemented using the structure of the block diagram of Figure 17) will be used in describing the flow chart of Figure 6. For example, modules may be stored in memory 1704 of Figure 17, and these modules may provide instructions so that when the instructions of a module are executed by respective network node processing circuitry 1702, processing circuitry 1702 directs the network node to perform respective operations of the flow chart.

[0069] Turning to Figure 6, in block 601, the network node 1700 transmits a plurality of reference signals on beams on a plurality of same time-frequency resources in all directions in accordance with power allocation vectors of a power allocation matrix.

[0070] In block 603, the network node 1700 receives, from a UE, composite signal strengths received for each power allocation vector transmitted.

[0071] In block 605, the network node 1700 calculates a sparse vector D using a compressed sensing algorithm based on a number of the composite signal strengths measured. In some embodiments, the network node 1700 calculates the sparse vector D based on a machine learning process. For example, the network node 1700 calculates the sparse vector D using one of an orthogonal matching pursue, OMP, an iterative hard thresholding, IHT, and a basis pursuit, BP. In other embodiments, the network node 1700 calculates the sparse vector D using the compressed sensing algorithm based on a number of the combined reference signals measured comprises iteratively calculating the sparse vector D using the compressed sensing algorithm based on different numbers of the combined reference signals measured.

[0072] In block 607, the network node 1700 determines best beams to use to transmit to the UE based on the sparse vector D. In some embodiments, the network node 1700 determines the best beam directions by determining which beam directions are above a threshold and indicating that the beam directions that are above the threshold are the best beam directions.

[0073] In some embodiments, the network node 1700 determines the power allocation matrix based on a machine learning process, wherein each row of the power allocation matrix defines a power allocation vector specifying how much power is allocated to each beam of a number of access points, APs, each AP having a number of beam directions per AP. In some of these embodiments, the network node 1700 determines the power allocation matrix based on dictionary learning.

[0074] In some embodiments of determining the power allocation based on dictionary learning, the network node 1700 collects learning measurement samples where the sparse vectors D are directly measured and determines an optimal power allocation matrix such that the sparse vectors D can be reconstructed from a shortest possible measurement vector.

[0075] In yet other embodiments, the network node 1700 determines the power allocation matrix based on dictionary learning by determining the power allocation matrix using an autoencoder neural network.

[0076] Uplink Rx side CS scanning

[0077] The embodiments described above assumes that the CS based signal combination is applied on the downlink Tx powers. In an alternative embodiment, it is possible to employ the CS signal combination in the uplink Rx beamforming such that the base station (BS) multiplies the multi-antenna received signals with different decoding vectors, which would create different reception beam patterns (corresponding to the rows of the CS matrix). The UE transmits in the uplink a sounding reference signal, which is coherently received at the BS(s) and are multiplied with the different decoding CS vectors. The UE needs to transmit as many sounding signal occasions as the number of CS reception combination.

[0078] In such an uplink application of the CS scheme, the measurement signals need to be sent and processed per UE (as opposed to the common reference signal in the downlink case), which may imply a larger reference signal overhead. However, it could be applied at initial access when the UE may send the request signal in a few times and the BS would try to receive the signal in a few beam decoding combinations. During the reception, the BS could receive the random access request and at the same time figure out the best beam direction (using the CS based scheme) and respond to the UE in the determined best beam direction.

[0079] Later, during data communication the per-UE uplink sounding reference signals would be sent periodically in-band and the BS may rotate the different beam decoding vectors (corresponding to the CS vectors) for the reception of these reference signals and based on that continuously determine the best beam direction.

[0080] Experimental results on real measurements

[0081] Initial testing using the embodiments described above using real measurements from a test network running on 39 GHz with 172 possible beam directions was done.

[0082] Figure 7 is a plot of the best beam index as a function of time as the UE moves around (the three lines correspond to three different routes).

[0083] Figure 8 is a plot of the number of beams that have measured signal strength above - 100 dBm. These are the beams that may provide measurable coverage to the UE. Note that in most of the time the number of beams with good enough signal strength is less than 10, which in comparison to the total number of 172 beams is small, indicating the sparsity of the possible beam directions.

[0084] A measurement matrix P (i.e., the power allocation matrix) can be created as a randomly generated matrix in the simplest case and one can perform compressed sensing based reconstruction of the sparse beam direction vector (D) from a few set of measurements. Recall, that each row of the dictionary matrix corresponds to one measurement combination (i.e., one power allocation combination). An example measurement matrix (i.e., power allocation dictionary) is shown in Figure 9.

[0085] In Figures 10A and 10B, the estimated beam directions (more specifically, the relative signal strength of the beams in linear scale) for the true measured values and for the reconstructed values (i.e., reconstructed from the few measurement combinations with the help of the CS algorithm) are shown. One can see that the true and estimated best beam directions nicely coincide, though there are some weaker "noisy" estimations in other directions as well, which practically have zero signal strength in the real measurements.

[0086] However, if the number of measurement combinations is slightly increased to 30, basically "perfect" beam direction estimations can be achieved, as shown in Figures 11 A and 11B.

[0087] Learning the dictionary - optimizing the measurements

[0088] The power allocation dictionary can be optimized to the particular radio environment, that is, to the typical beam constellation patterns prevailing in the given environment. The number of required measurements and transmission combinations can be significantly reduced if the dictionary is optimized. In the following describe possible dictionary learning solutions to optimize the power allocation matrix.

[0089] Note that the conventional dictionary learning algorithms available in compressed sensing are not applicable for this problem, as those algorithms are optimizing the dictionary and the sparse code at the same time, while in the beam direction case, the sparse coding should not be subject to optimization as that is “given” by the physical radio propagation environment.

[0090] To optimize the power allocation dictionary, it is assumed that a representative set of beam constellation patterns (sparse codes) are available for the algorithms. In practice, this dataset can be collected by active dedicated measurements (drive tests, automated AGVs, drones, etc.) or by building the database online, enabling continual learning. The latter option is feasible because the use case is valid and working with random power allocation dictionaries, while the optimized dictionaries make it more efficient with building up the database.

[0091] Learning by Deep Neural Network

[0092] In this embodiment the power allocation dictionary learning is performed with an autoencoder architecture such as an autoencoder Neural Network (NN). The dictionary is represented by the weights of a neural network layer in the encoder with linear activations and zero biases. The decoder NN module is used to perform the sparse coding on Y and the P dictionary, which is given explicitly to the decoder or learnt implicitly through a large training dataset (see Figure 12). The training of the autoencoder is the measured sparse code dataset (beam pattern). In some embodiments, the decoder NN module is implemented in a neural network-based sparse decoder.

[0093] Note that the decoder function and the loss function must be differentiable to be able to teach the NN with backpropagation. This can be achieved by a differentiable implementation of a sparse coding algorithm or this algorithm can be learnt by a NN module, which acts as a function approximator in this case, and it can be trained as part of the autoencoder or in advance. [0094] A further option for autoencoder implementation is to use the pseudo-inverse of dictionary weights P in the decoder module. In this case, Y is multiplied by pseudo-inv(P) to approximate the original input sparse code D. Figure 13 shows the resulting beam pattern estimation for 10 measurements.

[0095] Dictionary dimensionality reduction

[0096] In some embodiments, the set of sparse codes are projected into lower dimensional sub-space with dimensionality reduction techniques, like linear Principal Component Analysis (PCA), or any other suitable algorithm. The number dimensions of the sub-space Nfeatures can be defined in advance, which, eventually will determine the number of beam measurements for best beam selection. The subspace-search will provide a matrix P’ which is efficient in reducing the dimensions of the sparse codes and the Y values representing the aggregated measurements. By this, an efficient the power allocation matrix P can be derived as P = pseudo-inverse(P ) . A sample result of this implementation is shown in Figures 14A and 14B. Thus, the network node 1700 projects a set of sparse vector codes into a lower dimensional sub-space via one or more dimensionality reduction techniques and determines a power allocation sub-matrix in the lower dimensional space. The network node 1700 derives the power allocation matrix by determining a pseudo-inverse of the power allocation sub-matrix.

[0097] The proposed methods can be generally applied to any beamforming system, being it single AP or multiple AP system. When applied in a multi-AP scenario, e.g., in a D-MIMO setting, the benefits can be more significant than in a single point MIMO scenario as the number of possible beams or beam combinations are much higher in a D-MIMO setting. Moreover, the dictionary learning algorithm and the application of CS needs to be adapted to the D-MIMO case in a few details, as outlined below.

[0098] D-MIMO aspects

[0099] In the power allocation dictionary, the power of the beams corresponding to the same AP need to be normalized to the AP total transmission (Tx) power, while power to beams corresponding to different APs have independent sum of power constraints. This needs to be taken into account as constraints in the dictionary learning process, in particular in the cost function of the ML optimizer. Thus, the network node 1700 normalizes a power of beams corresponding to a same AP to AP total transmit power

[0100] Figure 15 shows an example of a communication system 1500 in accordance with some embodiments. [0101] In the example, the communication system 1500 includes a telecommunication network 1502 that includes an access network 1504, such as a radio access network (RAN), and a core network 1506, which includes one or more core network nodes 1508. The access network 1504 includes one or more access network nodes, such as network nodes 1510A and 1510B (one or more of which may be generally referred to as network nodes 1510), or any other similar 3^rd Generation Partnership Project (3 GPP) access node or non-3GPP access point. The network nodes 1510 facilitate direct or indirect connection of user equipment (UE), such as by connecting UEs 1512A, 1512B, 1512C, and 1512D (one or more of which may be generally referred to as UEs 1512) to the core network 1506 over one or more wireless connections.

[0102] Example wireless communications over a wireless connection include transmitting and/or receiving wireless signals using electromagnetic waves, radio waves, infrared waves, and/or other types of signals suitable for conveying information without the use of wires, cables, or other material conductors. Moreover, in different embodiments, the communication system 1500 may include any number of wired or wireless networks, network nodes, UEs, and/or any other components or systems that may facilitate or participate in the communication of data and/or signals whether via wired or wireless connections. The communication system 1500 may include and/or interface with any type of communication, telecommunication, data, cellular, radio network, and/or other similar type of system.

[0103] The UEs 1512 may be any of a wide variety of communication devices, including wireless devices arranged, configured, and/or operable to communicate wirelessly with the network nodes 1510 and other communication devices. Similarly, the network nodes 1510 are arranged, capable, configured, and/or operable to communicate directly or indirectly with the UEs 1512 and/or with other network nodes or equipment in the telecommunication network 1502 to enable and/or provide network access, such as wireless network access, and/or to perform other functions, such as administration in the telecommunication network 1502.

[0104] In the depicted example, the core network 1506 connects the network nodes 1510 to one or more hosts, such as host 1516. These connections may be direct or indirect via one or more intermediary networks or devices. In other examples, network nodes may be directly coupled to hosts. The core network 1506 includes one more core network nodes (e.g., core network node 1508) that are structured with hardware and software components. Features of these components may be substantially similar to those described with respect to the UEs, network nodes, and/or hosts, such that the descriptions thereof are generally applicable to the corresponding components of the core network node 1508. Example core network nodes include functions of one or more of a Mobile Switching Center (MSC), Mobility Management Entity (MME), Home Subscriber Server (HSS), Access and Mobility Management Function (AMF), Session Management Function (SMF), Authentication Server Function (AUSF), Subscription Identifier De-concealing function (SIDF), Unified Data Management (UDM), Security Edge Protection Proxy (SEPP), Network Exposure Function (NEF), and/or a User Plane Function (UPF).

[0105] The host 1516 may be under the ownership or control of a service provider other than an operator or provider of the access network 1504 and/or the telecommunication network 1502, and may be operated by the service provider or on behalf of the service provider. The host 1516 may host a variety of applications to provide one or more service. Examples of such applications include live and pre-recorded audio/video content, data collection services such as retrieving and compiling data on various ambient conditions detected by a plurality of UEs, analytics functionality, social media, functions for controlling or otherwise interacting with remote devices, functions for an alarm and surveillance center, or any other such function performed by a server.

[0106] As a whole, the communication system 1500 of Figure 15 enables connectivity between the UEs, network nodes, and hosts. In that sense, the communication system may be configured to operate according to predefined rules or procedures, such as specific standards that include, but are not limited to: Global System for Mobile Communications (GSM); Universal Mobile Telecommunications System (UMTS); Long Term Evolution (LTE), and/or other suitable 2G, 3G, 4G, 5G standards, or any applicable future generation standard (e.g., 6G); wireless local area network (WLAN) standards, such as the Institute of Electrical and Electronics Engineers (IEEE) 802.11 standards (WiFi); and/or any other appropriate wireless communication standard, such as the Worldwide Interoperability for Microwave Access (WiMax), Bluetooth, Z- Wave, Near Field Communication (NFC) ZigBee, LiFi, and/or any low-power wide-area network (LPWAN) standards such as LoRa and Sigfox.

[0107] In some examples, the telecommunication network 1502 is a cellular network that implements 3GPP standardized features. Accordingly, the telecommunications network 1502 may support network slicing to provide different logical networks to different devices that are connected to the telecommunication network 1502. For example, the telecommunications network 1502 may provide Ultra Reliable Low Latency Communication (URLLC) services to some UEs, while providing Enhanced Mobile Broadband (eMBB) services to other UEs, and/or Massive Machine Type Communication (mMTC)ZMassive loT services to yet further UEs.

[0108] In some examples, the UEs 1512 are configured to transmit and/or receive information without direct human interaction. For instance, a UE may be designed to transmit information to the access network 1504 on a predetermined schedule, when triggered by an internal or external event, or in response to requests from the access network 1504. Additionally, a UE may be configured for operating in single- or multi-RAT or multi -standard mode. For example, a UE may operate with any one or combination of Wi-Fi, NR (New Radio) and LTE, i.e., being configured for multi-radio dual connectivity (MR-DC), such as E-UTRAN (Evolved- UMTS Terrestrial Radio Access Network) New Radio - Dual Connectivity (EN-DC).

[0109] In the example, the hub 1514 communicates with the access network 1504 to facilitate indirect communication between one or more UEs (e.g., UE 1512C and/or 1512D) and network nodes (e.g., network node 1510B). In some examples, the hub 1514 may be a controller, router, content source and analytics, or any of the other communication devices described herein regarding UEs. For example, the hub 1514 may be a broadband router enabling access to the core network 1506 for the UEs. As another example, the hub 1514 may be a controller that sends commands or instructions to one or more actuators in the UEs. Commands or instructions may be received from the UEs, network nodes 1510, or by executable code, script, process, or other instructions in the hub 1514. As another example, the hub 1514 may be a data collector that acts as temporary storage for UE data and, in some embodiments, may perform analysis or other processing of the data. As another example, the hub 1514 may be a content source. For example, for a UE that is a VR headset, display, loudspeaker or other media delivery device, the hub 1514 may retrieve VR assets, video, audio, or other media or data related to sensory information via a network node, which the hub 1514 then provides to the UE either directly, after performing local processing, and/or after adding additional local content. In still another example, the hub 1514 acts as a proxy server or orchestrator for the UEs, in particular in if one or more of the UEs are low energy loT devices.

[0110] The hub 1514 may have a constant/persistent or intermittent connection to the network node 1510B. The hub 1514 may also allow for a different communication scheme and/or schedule between the hub 1514 and UEs (e.g., UE 1512C and/or 1512D), and between the hub 1514 and the core network 1506. In other examples, the hub 1514 is connected to the core network 1506 and/or one or more UEs via a wired connection. Moreover, the hub 1514 may be configured to connect to an M2M service provider over the access network 1504 and/or to another UE over a direct connection. In some scenarios, UEs may establish a wireless connection with the network nodes 1510 while still connected via the hub 1514 via a wired or wireless connection. In some embodiments, the hub 1514 may be a dedicated hub - that is, a hub whose primary function is to route communications to/from the UEs from/to the network node 1510B. In other embodiments, the hub 1514 may be a non-dedicated hub - that is, a device which is capable of operating to route communications between the UEs and network node 1510B, but which is additionally capable of operating as a communication start and/or end point for certain data channels.

[0111] Figure 16 shows a UE 1600 in accordance with some embodiments. As used herein, a UE refers to a device capable, configured, arranged and/or operable to communicate wirelessly with network nodes and/or other UEs. Examples of a UE include, but are not limited to, a smart phone, mobile phone, cell phone, voice over IP (VoIP) phone, wireless local loop phone, desktop computer, personal digital assistant (PDA), wireless cameras, gaming console or device, music storage device, playback appliance, wearable terminal device, wireless endpoint, mobile station, tablet, laptop, laptop-embedded equipment (LEE), laptop-mounted equipment (LME), smart device, wireless customer-premise equipment (CPE), vehicle-mounted or vehicle embedded/integrated wireless device, etc. Other examples include any UE identified by the 3rd Generation Partnership Project (3GPP), including a narrow band internet of things (NB-IoT) UE, a machine type communication (MTC) UE, and/or an enhanced MTC (eMTC) UE.

[0112] A UE may support device-to-device (D2D) communication, for example by implementing a 3 GPP standard for sidelink communication, Dedicated Short-Range Communication (DSRC), vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), or vehicle- to-everything (V2X). In other examples, a UE may not necessarily have a user in the sense of a human user who owns and/or operates the relevant device. Instead, a UE may represent a device that is intended for sale to, or operation by, a human user but which may not, or which may not initially, be associated with a specific human user (e.g., a smart sprinkler controller).

Alternatively, a UE may represent a device that is not intended for sale to, or operation by, an end user but which may be associated with or operated for the benefit of a user (e.g., a smart power meter).

[0113] The UE 1600 includes processing circuitry 1602 that is operatively coupled via a bus 1604 to an input/output interface 1606, a power source 1608, a memory 1610, a communication interface 1612, and/or any other component, or any combination thereof. Certain UEs may utilize all or a subset of the components shown in Figure 16. The level of integration between the components may vary from one UE to another UE. Further, certain UEs may contain multiple instances of a component, such as multiple processors, memories, transceivers, transmitters, receivers, etc.

[0114] The processing circuitry 1602 is configured to process instructions and data and may be configured to implement any sequential state machine operative to execute instructions stored as machine-readable computer programs in the memory 1610. The processing circuitry 1602 may be implemented as one or more hardware-implemented state machines (e.g., in discrete logic, field-programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), etc.); programmable logic together with appropriate firmware; one or more stored computer programs, general-purpose processors, such as a microprocessor or digital signal processor (DSP), together with appropriate software; or any combination of the above. For example, the processing circuitry 1602 may include multiple central processing units (CPUs).

[0115] In the example, the input/output interface 1606 may be configured to provide an interface or interfaces to an input device, output device, or one or more input and/or output devices. Examples of an output device include a speaker, a sound card, a video card, a display, a monitor, a printer, an actuator, an emitter, a smartcard, another output device, or any combination thereof. An input device may allow a user to capture information into the UE 1600. Examples of an input device include a touch-sensitive or presence-sensitive display, a camera (e.g., a digital camera, a digital video camera, a web camera, etc.), a microphone, a sensor, a mouse, a trackball, a directional pad, a trackpad, a scroll wheel, a smartcard, and the like. The presence-sensitive display may include a capacitive or resistive touch sensor to sense input from a user. A sensor may be, for instance, an accelerometer, a gyroscope, a tilt sensor, a force sensor, a magnetometer, an optical sensor, a proximity sensor, a biometric sensor, etc., or any combination thereof. An output device may use the same type of interface port as an input device. For example, a Universal Serial Bus (USB) port may be used to provide an input device and an output device.

[0116] In some embodiments, the power source 1608 is structured as a battery or battery pack. Other types of power sources, such as an external power source (e.g., an electricity outlet), photovoltaic device, or power cell, may be used. The power source 1608 may further include power circuitry for delivering power from the power source 1608 itself, and/or an external power source, to the various parts of the UE 1600 via input circuitry or an interface such as an electrical power cable. Delivering power may be, for example, for charging of the power source 1608. Power circuitry may perform any formatting, converting, or other modification to the power from the power source 1608 to make the power suitable for the respective components of the UE 1600 to which power is supplied.

[0117] The memory 1610 may be or be configured to include memory such as random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable readonly memory (EEPROM), magnetic disks, optical disks, hard disks, removable cartridges, flash drives, and so forth. In one example, the memory 1610 includes one or more application programs 1614, such as an operating system, web browser application, a widget, gadget engine, or other application, and corresponding data 1616. The memory 1610 may store, for use by the UE 1600, any of a variety of various operating systems or combinations of operating systems. [0118] The memory 1610 may be configured to include a number of physical drive units, such as redundant array of independent disks (RAID), flash memory, USB flash drive, external hard disk drive, thumb drive, pen drive, key drive, high-density digital versatile disc (HD-DVD) optical disc drive, internal hard disk drive, Blu-Ray optical disc drive, holographic digital data storage (HDDS) optical disc drive, external mini-dual in-line memory module (DIMM), synchronous dynamic random access memory (SDRAM), external micro-DIMM SDRAM, smartcard memory such as tamper resistant module in the form of a universal integrated circuit card (UICC) including one or more subscriber identity modules (SIMs), such as a USIM and/or ISIM, other memory, or any combination thereof. The UICC may for example be an embedded UICC (eUICC), integrated UICC (iUICC) or a removable UICC commonly known as ‘ SIM card.’ The memory 1610 may allow the UE 1600 to access instructions, application programs and the like, stored on transitory or non-transitory memory media, to off-load data, or to upload data. An article of manufacture, such as one utilizing a communication system may be tangibly embodied as or in the memory 1610, which may be or comprise a device-readable storage medium.

[0119] The processing circuitry 1602 may be configured to communicate with an access network or other network using the communication interface 1612. The communication interface 1612 may comprise one or more communication subsystems and may include or be communicatively coupled to an antenna 1622. The communication interface 1612 may include one or more transceivers used to communicate, such as by communicating with one or more remote transceivers of another device capable of wireless communication (e.g., another UE or a network node in an access network). Each transceiver may include a transmitter 1618 and/or a receiver 1620 appropriate to provide network communications (e.g., optical, electrical, frequency allocations, and so forth). Moreover, the transmitter 1618 and receiver 1620 may be coupled to one or more antennas (e.g., antenna 1622) and may share circuit components, software or firmware, or alternatively be implemented separately.

[0120] In the illustrated embodiment, communication functions of the communication interface 1612 may include cellular communication, Wi-Fi communication, LPWAN communication, data communication, voice communication, multimedia communication, short- range communications such as Bluetooth, near-field communication, location-based communication such as the use of the global positioning system (GPS) to determine a location, another like communication function, or any combination thereof. Communications may be implemented in according to one or more communication protocols and/or standards, such as IEEE 802.11, Code Division Multiplexing Access (CDMA), Wideband Code Division Multiple Access (WCDMA), GSM, LTE, New Radio (NR), UMTS, WiMax, Ethernet, transmission control protocol/intemet protocol (TCP/IP), synchronous optical networking (SONET), Asynchronous Transfer Mode (ATM), QUIC, Hypertext Transfer Protocol (HTTP), and so forth. [0121] Regardless of the type of sensor, a UE may provide an output of data captured by its sensors, through its communication interface 1612, via a wireless connection to a network node. Data captured by sensors of a UE can be communicated through a wireless connection to a network node via another UE. The output may be periodic (e.g., once every 15 minutes if it reports the sensed temperature), random (e.g., to even out the load from reporting from several sensors), in response to a triggering event (e.g., when moisture is detected an alert is sent), in response to a request (e.g., a user initiated request), or a continuous stream (e.g., a live video feed of a patient).

[0122] As another example, a UE comprises an actuator, a motor, or a switch, related to a communication interface configured to receive wireless input from a network node via a wireless connection. In response to the received wireless input the states of the actuator, the motor, or the switch may change. For example, the UE may comprise a motor that adjusts the control surfaces or rotors of a drone in flight according to the received input or to a robotic arm performing a medical procedure according to the received input.

[0123] A UE, when in the form of an Internet of Things (loT) device, may be a device for use in one or more application domains, these domains comprising, but not limited to, city wearable technology, extended industrial application and healthcare. Non-limiting examples of such an loT device are a device which is or which is embedded in: a connected refrigerator or freezer, a TV, a connected lighting device, an electricity meter, a robot vacuum cleaner, a voice controlled smart speaker, a home security camera, a motion detector, a thermostat, a smoke detector, a door/window sensor, a flood/moisture sensor, an electrical door lock, a connected doorbell, an air conditioning system like a heat pump, an autonomous vehicle, a surveillance system, a weather monitoring device, a vehicle parking monitoring device, an electric vehicle charging station, a smart watch, a fitness tracker, a head-mounted display for Augmented Reality (AR) or Virtual Reality (VR), a wearable for tactile augmentation or sensory enhancement, a water sprinkler, an animal- or item-tracking device, a sensor for monitoring a plant or animal, an industrial robot, an Unmanned Aerial Vehicle (UAV), and any kind of medical device, like a heart rate monitor or a remote controlled surgical robot. A UE in the form of an loT device comprises circuitry and/or software in dependence of the intended application of the loT device in addition to other components as described in relation to the UE 1600 shown in Figure 16. [0124] As yet another specific example, in an loT scenario, a UE may represent a machine or other device that performs monitoring and/or measurements, and transmits the results of such monitoring and/or measurements to another UE and/or a network node. The UE may in this case be an M2M device, which may in a 3GPP context be referred to as an MTC device. As one particular example, the UE may implement the 3GPP NB-IoT standard. In other scenarios, a UE may represent a vehicle, such as a car, a bus, a truck, a ship and an airplane, or other equipment that is capable of monitoring and/or reporting on its operational status or other functions associated with its operation.

[0125] In practice, any number of UEs may be used together with respect to a single use case. For example, a first UE might be or be integrated in a drone and provide the drone’s speed information (obtained through a speed sensor) to a second UE that is a remote controller operating the drone. When the user makes changes from the remote controller, the first UE may adjust the throttle on the drone (e.g., by controlling an actuator) to increase or decrease the drone’s speed. The first and/or the second UE can also include more than one of the functionalities described above. For example, a UE might comprise the sensor and the actuator, and handle communication of data for both the speed sensor and the actuators.

[0126] Figure 17 shows a network node 1700 in accordance with some embodiments. As used herein, network node refers to equipment capable, configured, arranged and/or operable to communicate directly or indirectly with a UE and/or with other network nodes or equipment, in a telecommunication network. Examples of network nodes include, but are not limited to, access points (APs) (e.g., radio access points), base stations (BSs) (e.g., radio base stations, Node Bs, evolved Node Bs (eNBs) and NR. NodeBs (gNBs)).

[0127] Base stations may be categorized based on the amount of coverage they provide (or, stated differently, their transmit power level) and so, depending on the provided amount of coverage, may be referred to as femto base stations, pico base stations, micro base stations, or macro base stations. A base station may be a relay node or a relay donor node controlling a relay. A network node may also include one or more (or all) parts of a distributed radio base station such as centralized digital units and/or remote radio units (RRUs), sometimes referred to as Remote Radio Heads (RRHs). Such remote radio units may or may not be integrated with an antenna as an antenna integrated radio. Parts of a distributed radio base station may also be referred to as nodes in a distributed antenna system (DAS).

[0128] Other examples of network nodes include multiple transmission point (multi-TRP) 5G access nodes, multi -standard radio (MSR) equipment such as MSR BSs, network controllers such as radio network controllers (RNCs) or base station controllers (BSCs), base transceiver stations (BTSs), transmission points, transmission nodes, multi -cell/multicast coordination entities (MCEs), Operation and Maintenance (O&M) nodes, Operations Support System (OSS) nodes, Self-Organizing Network (SON) nodes, positioning nodes (e.g., Evolved Serving Mobile Location Centers (E-SMLCs)), and/or Minimization of Drive Tests (MDTs).

[0129] The network node 1700 includes a processing circuitry 1702, a memory 1704, a communication interface 1706, and a power source 1708. The network node 1700 may be composed of multiple physically separate components (e.g., a NodeB component and a RNC component, or a BTS component and a BSC component, etc.), which may each have their own respective components. In certain scenarios in which the network node 1700 comprises multiple separate components (e.g., BTS and BSC components), one or more of the separate components may be shared among several network nodes. For example, a single RNC may control multiple NodeBs. In such a scenario, each unique NodeB and RNC pair, may in some instances be considered a single separate network node. In some embodiments, the network node 1700 may be configured to support multiple radio access technologies (RATs). In such embodiments, some components may be duplicated (e.g., separate memory 1704 for different RATs) and some components may be reused (e.g., a same antenna 1710 may be shared by different RATs). The network node 1700 may also include multiple sets of the various illustrated components for different wireless technologies integrated into network node 1700, for example GSM, WCDMA, LTE, NR, WiFi, Zigbee, Z-wave, LoRaWAN, Radio Frequency Identification (RFID) or Bluetooth wireless technologies. These wireless technologies may be integrated into the same or different chip or set of chips and other components within network node 1700.

[0130] The processing circuitry 1702 may comprise a combination of one or more of a microprocessor, controller, microcontroller, central processing unit, digital signal processor, application-specific integrated circuit, field programmable gate array, or any other suitable computing device, resource, or combination of hardware, software and/or encoded logic operable to provide, either alone or in conjunction with other network node 1700 components, such as the memory 1704, to provide network node 1700 functionality.

[0131] In some embodiments, the processing circuitry 1702 includes a system on a chip (SOC). In some embodiments, the processing circuitry 1702 includes one or more of radio frequency (RF) transceiver circuitry 1712 and baseband processing circuitry 1714. In some embodiments, the radio frequency (RF) transceiver circuitry 1712 and the baseband processing circuitry 1714 may be on separate chips (or sets of chips), boards, or units, such as radio units and digital units. In alternative embodiments, part or all of RF transceiver circuitry 1712 and baseband processing circuitry 1714 may be on the same chip or set of chips, boards, or units. [0132] The memory 1704 may comprise any form of volatile or non-volatile computer- readable memory including, without limitation, persistent storage, solid-state memory, remotely mounted memory, magnetic media, optical media, random access memory (RAM), read-only memory (ROM), mass storage media (for example, a hard disk), removable storage media (for example, a flash drive, a Compact Disk (CD) or a Digital Video Disk (DVD)), and/or any other volatile or non-volatile, non-transitory device-readable and/or computer-executable memory devices that store information, data, and/or instructions that may be used by the processing circuitry 1702. The memory 1704 may store any suitable instructions, data, or information, including a computer program, software, an application including one or more of logic, rules, code, tables, and/or other instructions capable of being executed by the processing circuitry 1702 and utilized by the network node 1700. The memory 1704 may be used to store any calculations made by the processing circuitry 1702 and/or any data received via the communication interface 1706. In some embodiments, the processing circuitry 1702 and memory 1704 is integrated.

[0133] The communication interface 1706 is used in wired or wireless communication of signaling and/or data between a network node, access network, and/or UE. As illustrated, the communication interface 1706 comprises port(s)/terminal(s) 1716 to send and receive data, for example to and from a network over a wired connection. The communication interface 1706 also includes radio front-end circuitry 1718 that may be coupled to, or in certain embodiments a part of, the antenna 1710. Radio front-end circuitry 1718 comprises filters 1720 and amplifiers 1722. The radio front-end circuitry 1718 may be connected to an antenna 1710 and processing circuitry 1702. The radio front-end circuitry may be configured to condition signals communicated between antenna 1710 and processing circuitry 1702. The radio front-end circuitry 1718 may receive digital data that is to be sent out to other network nodes or UEs via a wireless connection. The radio front-end circuitry 1718 may convert the digital data into a radio signal having the appropriate channel and bandwidth parameters using a combination of filters 1720 and/or amplifiers 1722. The radio signal may then be transmitted via the antenna 1710. Similarly, when receiving data, the antenna 1710 may collect radio signals which are then converted into digital data by the radio front-end circuitry 1718. The digital data may be passed to the processing circuitry 1702. In other embodiments, the communication interface may comprise different components and/or different combinations of components.

[0134] In certain alternative embodiments, the network node 1700 does not include separate radio front-end circuitry 1718, instead, the processing circuitry 1702 includes radio front-end circuitry and is connected to the antenna 1710. Similarly, in some embodiments, all or some of the RF transceiver circuitry 1712 is part of the communication interface 1706. In still other embodiments, the communication interface 1706 includes one or more ports or terminals 1716, the radio front-end circuitry 1718, and the RF transceiver circuitry 1712, as part of a radio unit (not shown), and the communication interface 1706 communicates with the baseband processing circuitry 1714, which is part of a digital unit (not shown).

[0135] The antenna 1710 may include one or more antennas, or antenna arrays, configured to send and/or receive wireless signals. The antenna 1710 may be coupled to the radio front-end circuitry 1718 and may be any type of antenna capable of transmitting and receiving data and/or signals wirelessly. In certain embodiments, the antenna 1710 is separate from the network node 1700 and connectable to the network node 1700 through an interface or port.

[0136] The antenna 1710, communication interface 1706, and/or the processing circuitry 1702 may be configured to perform any receiving operations and/or certain obtaining operations described herein as being performed by the network node. Any information, data and/or signals may be received from a UE, another network node and/or any other network equipment. Similarly, the antenna 1710, the communication interface 1706, and/or the processing circuitry 1702 may be configured to perform any transmitting operations described herein as being performed by the network node. Any information, data and/or signals may be transmitted to a UE, another network node and/or any other network equipment.

[0137] The power source 1708 provides power to the various components of network node 1700 in a form suitable for the respective components (e.g., at a voltage and current level needed for each respective component). The power source 1708 may further comprise, or be coupled to, power management circuitry to supply the components of the network node 1700 with power for performing the functionality described herein. For example, the network node 1700 may be connectable to an external power source (e.g., the power grid, an electricity outlet) via an input circuitry or interface such as an electrical cable, whereby the external power source supplies power to power circuitry of the power source 1708. As a further example, the power source 1708 may comprise a source of power in the form of a battery or battery pack which is connected to, or integrated in, power circuitry. The battery may provide backup power should the external power source fail.

[0138] Embodiments of the network node 1700 may include additional components beyond those shown in Figure 17 for providing certain aspects of the network node’s functionality, including any of the functionality described herein and/or any functionality necessary to support the subject matter described herein. For example, the network node 1700 may include user interface equipment to allow input of information into the network node 1700 and to allow output of information from the network node 1700. This may allow a user to perform diagnostic, maintenance, repair, and other administrative functions for the network node 1700.

[0139] Figure 18 is a block diagram of a host 1800, which may be an embodiment of the host 1516 of Figure 15, in accordance with various aspects described herein. As used herein, the host 1800 may be or comprise various combinations hardware and/or software, including a standalone server, a blade server, a cloud-implemented server, a distributed server, a virtual machine, container, or processing resources in a server farm. The host 1800 may provide one or more services to one or more UEs.

[0140] The host 1800 includes processing circuitry 1802 that is operatively coupled via a bus 1804 to an input/output interface 1806, a network interface 1808, a power source 1810, and a memory 1812. Other components may be included in other embodiments. Features of these components may be substantially similar to those described with respect to the devices of previous figures, such as Figures 16 and 17, such that the descriptions thereof are generally applicable to the corresponding components of host 1800.

[0141] The memory 1812 may include one or more computer programs including one or more host application programs 1814 and data 1816, which may include user data, e.g., data generated by a UE for the host 1800 or data generated by the host 1800 for a UE. Embodiments of the host 1800 may utilize only a subset or all of the components shown. The host application programs 1814 may be implemented in a container-based architecture and may provide support for video codecs (e.g., Versatile Video Coding (VVC), High Efficiency Video Coding (HEVC), Advanced Video Coding (AVC), MPEG, VP9) and audio codecs (e.g., FLAC, Advanced Audio Coding (AAC), MPEG, G.711), including transcoding for multiple different classes, types, or implementations of UEs (e.g., handsets, desktop computers, wearable display systems, heads-up display systems). The host application programs 1814 may also provide for user authentication and licensing checks and may periodically report health, routes, and content availability to a central node, such as a device in or on the edge of a core network. Accordingly, the host 1800 may select and/or indicate a different host for over-the-top services for a UE. The host application programs 1814 may support various protocols, such as the HTTP Live Streaming (HLS) protocol, Real-Time Messaging Protocol (RTMP), Real-Time Streaming Protocol (RTSP), Dynamic Adaptive Streaming over HTTP (MPEG-DASH), etc.

[0142] Figure 19 is a block diagram illustrating a virtualization environment 1900 in which functions implemented by some embodiments may be virtualized. In the present context, virtualizing means creating virtual versions of apparatuses or devices which may include virtualizing hardware platforms, storage devices and networking resources. As used herein, virtualization can be applied to any device described herein, or components thereof, and relates to an implementation in which at least a portion of the functionality is implemented as one or more virtual components. Some or all of the functions described herein may be implemented as virtual components executed by one or more virtual machines (VMs) implemented in one or more virtual environments 1900 hosted by one or more of hardware nodes, such as a hardware computing device that operates as a network node, UE, core network node, or host. Further, in embodiments in which the virtual node does not require radio connectivity (e.g., a core network node or host), then the node may be entirely virtualized.

[0143] Applications 1902 (which may alternatively be called software instances, virtual appliances, network functions, virtual nodes, virtual network functions, etc.) are run in the virtualization environment Q400 to implement some of the features, functions, and/or benefits of some of the embodiments disclosed herein.

[0144] Hardware 1904 includes processing circuitry, memory that stores software and/or instructions executable by hardware processing circuitry, and/or other hardware devices as described herein, such as a network interface, input/output interface, and so forth. Software may be executed by the processing circuitry to instantiate one or more virtualization layers 1906 (also referred to as hypervisors or virtual machine monitors (VMMs)), provide VMs 1908a and 1908b (one or more of which may be generally referred to as VMs 1908), and/or perform any of the functions, features and/or benefits described in relation with some embodiments described herein. The virtualization layer 1906 may present a virtual operating platform that appears like networking hardware to the VMs 1908.

[0145] The VMs 1908 comprise virtual processing, virtual memory, virtual networking or interface and virtual storage, and may be run by a corresponding virtualization layer 1906. Different embodiments of the instance of a virtual appliance 1902 may be implemented on one or more of VMs 1908, and the implementations may be made in different ways. Virtualization of the hardware is in some contexts referred to as network function virtualization (NFV). NFV may be used to consolidate many network equipment types onto industry standard high volume server hardware, physical switches, and physical storage, which can be located in data centers, and customer premise equipment.

[0146] In the context of NFV, a VM 1908 may be a software implementation of a physical machine that runs programs as if they were executing on a physical, non-virtualized machine. Each of the VMs 1908, and that part of hardware 1904 that executes that VM, be it hardware dedicated to that VM and/or hardware shared by that VM with others of the VMs, forms separate virtual network elements. Still in the context of NFV, a virtual network function is responsible for handling specific network functions that run in one or more VMs 1908 on top of the hardware 1904 and corresponds to the application 1902.

[0147] Hardware 1904 may be implemented in a standalone network node with generic or specific components. Hardware 1904 may implement some functions via virtualization.

Alternatively, hardware 1904 may be part of a larger cluster of hardware (e.g., such as in a data center or CPE) where many hardware nodes work together and are managed via management and orchestration 1910, which, among others, oversees lifecycle management of applications 1902. In some embodiments, hardware 1904 is coupled to one or more radio units that each include one or more transmitters and one or more receivers that may be coupled to one or more antennas. Radio units may communicate directly with other hardware nodes via one or more appropriate network interfaces and may be used in combination with the virtual components to provide a virtual node with radio capabilities, such as a radio access node or a base station. In some embodiments, some signaling can be provided with the use of a control system 1912 which may alternatively be used for communication between hardware nodes and radio units.

[0148] Figure 20 shows a communication diagram of a host 2002 communicating via a network node 2004 with a UE 2006 over a partially wireless connection in accordance with some embodiments. Example implementations, in accordance with various embodiments, of the UE (such as a UE 1512A of Figure 15 and/or UE 1600 of Figure 16), network node (such as network node 1510a of Figure 15 and/or network node 1700 of Figure 17), and host (such as host 1516 of Figure 15 and/or host 1800 of Figure 18) discussed in the preceding paragraphs will now be described with reference to Figure 20.

[0149] Like host 1800, embodiments of host 2002 include hardware, such as a communication interface, processing circuitry, and memory. The host 2002 also includes software, which is stored in or accessible by the host 2002 and executable by the processing circuitry. The software includes a host application that may be operable to provide a service to a remote user, such as the UE 2006 connecting via an over-the-top (OTT) connection 2050 extending between the UE 2006 and host 2002. In providing the service to the remote user, a host application may provide user data which is transmitted using the OTT connection 2050. [0150] The network node 2004 includes hardware enabling it to communicate with the host 2002 and UE 2006. The connection 2060 may be direct or pass through a core network (like core network 1506 of Figure 15) and/or one or more other intermediate networks, such as one or more public, private, or hosted networks. For example, an intermediate network may be a backbone network or the Internet. [0151] The UE 2006 includes hardware and software, which is stored in or accessible by UE 2006 and executable by the UE’s processing circuitry. The software includes a client application, such as a web browser or operator-specific “app” that may be operable to provide a service to a human or non-human user via UE 2006 with the support of the host 2002. In the host 2002, an executing host application may communicate with the executing client application via the OTT connection 2050 terminating at the UE 2006 and host 2002. In providing the service to the user, the UE's client application may receive request data from the host's host application and provide user data in response to the request data. The OTT connection 2050 may transfer both the request data and the user data. The UE's client application may interact with the user to generate the user data that it provides to the host application through the OTT connection 2050. [0152] The OTT connection 2050 may extend via a connection 2060 between the host 2002 and the network node 2004 and via a wireless connection 2070 between the network node 2004 and the UE 2006 to provide the connection between the host 2002 and the UE 2006. The connection 2060 and wireless connection 2070, over which the OTT connection 2050 may be provided, have been drawn abstractly to illustrate the communication between the host 2002 and the UE 2006 via the network node 2004, without explicit reference to any intermediary devices and the precise routing of messages via these devices.

[0153] As an example of transmitting data via the OTT connection 2050, in step 2008, the host 2002 provides user data, which may be performed by executing a host application. In some embodiments, the user data is associated with a particular human user interacting with the UE 2006. In other embodiments, the user data is associated with a UE 2006 that shares data with the host 2002 without explicit human interaction. In step 2010, the host 2002 initiates a transmission carrying the user data towards the UE 2006. The host 2002 may initiate the transmission responsive to a request transmitted by the UE 2006. The request may be caused by human interaction with the UE 2006 or by operation of the client application executing on the UE 2006. The transmission may pass via the network node 2004, in accordance with the teachings of the embodiments described throughout this disclosure. Accordingly, in step 2012, the network node 2004 transmits to the UE 2006 the user data that was carried in the transmission that the host 2002 initiated, in accordance with the teachings of the embodiments described throughout this disclosure. In step 2014, the UE 2006 receives the user data carried in the transmission, which may be performed by a client application executed on the UE 2006 associated with the host application executed by the host 2002.

[0154] In some examples, the UE 2006 executes a client application which provides user data to the host 2002. The user data may be provided in reaction or response to the data received from the host 2002. Accordingly, in step 2016, the UE 2006 may provide user data, which may be performed by executing the client application. In providing the user data, the client application may further consider user input received from the user via an input/output interface of the UE 2006. Regardless of the specific manner in which the user data was provided, the UE 2006 initiates, in step 2018, transmission of the user data towards the host 2002 via the network node 2004. In step 2020, in accordance with the teachings of the embodiments described throughout this disclosure, the network node 2004 receives user data from the UE 2006 and initiates transmission of the received user data towards the host 2002. In step 2022, the host 2002 receives the user data carried in the transmission initiated by the UE 2006.

[0155] One or more of the various embodiments may improve the performance of OTT services provided to the UE 2006 using the OTT connection 2050, in which the wireless connection 2070 forms the last segment.

[0156] In an example scenario, factory status information may be collected and analyzed by the host 2002. As another example, the host 2002 may process audio and video data which may have been retrieved from a UE for use in creating maps. As another example, the host 2002 may collect and analyze real-time data to assist in controlling vehicle congestion (e.g., controlling traffic lights). As another example, the host 2002 may store surveillance video uploaded by a UE. As another example, the host 2002 may store or control access to media content such as video, audio, VR or AR which it can broadcast, multicast or unicast to UEs. As other examples, the host 2002 may be used for energy pricing, remote control of non-time critical electrical load to balance power generation needs, location services, presentation services (such as compiling diagrams etc. from data collected from remote devices), or any other function of collecting, retrieving, storing, analyzing and/or transmitting data.

[0157] In some examples, a measurement procedure may be provided for the purpose of monitoring data rate, latency and other factors on which the one or more embodiments improve. There may further be an optional network functionality for reconfiguring the OTT connection 2050 between the host 2002 and UE 2006, in response to variations in the measurement results. The measurement procedure and/or the network functionality for reconfiguring the OTT connection may be implemented in software and hardware of the host 2002 and/or UE 2006. In some embodiments, sensors (not shown) may be deployed in or in association with other devices through which the OTT connection 2050 passes; the sensors may participate in the measurement procedure by supplying values of the monitored quantities exemplified above, or supplying values of other physical quantities from which software may compute or estimate the monitored quantities. The reconfiguring of the OTT connection 2050 may include message format, retransmission settings, preferred routing etc.; the reconfiguring need not directly alter the operation of the network node 2004. Such procedures and functionalities may be known and practiced in the art. In certain embodiments, measurements may involve proprietary UE signaling that facilitates measurements of throughput, propagation times, latency and the like, by the host 2002. The measurements may be implemented in that software causes messages to be transmitted, in particular empty or ‘dummy’ messages, using the OTT connection 2050 while monitoring propagation times, errors, etc.

[0158] Although the computing devices described herein (e.g., UEs, network nodes, hosts) may include the illustrated combination of hardware components, other embodiments may comprise computing devices with different combinations of components. It is to be understood that these computing devices may comprise any suitable combination of hardware and/or software needed to perform the tasks, features, functions and methods disclosed herein. Determining, calculating, obtaining or similar operations described herein may be performed by processing circuitry, which may process information by, for example, converting the obtained information into other information, comparing the obtained information or converted information to information stored in the network node, and/or performing one or more operations based on the obtained information or converted information, and as a result of said processing making a determination. Moreover, while components are depicted as single boxes located within a larger box, or nested within multiple boxes, in practice, computing devices may comprise multiple different physical components that make up a single illustrated component, and functionality may be partitioned between separate components. For example, a communication interface may be configured to include any of the components described herein, and/or the functionality of the components may be partitioned between the processing circuitry and the communication interface. In another example, non-computationally intensive functions of any of such components may be implemented in software or firmware and computationally intensive functions may be implemented in hardware.

[0159] In certain embodiments, some or all of the functionality described herein may be provided by processing circuitry executing instructions stored on in memory, which in certain embodiments may be a computer program product in the form of a non-transitory computer- readable storage medium. In alternative embodiments, some or all of the functionality may be provided by the processing circuitry without executing instructions stored on a separate or discrete device-readable storage medium, such as in a hard-wired manner. In any of those particular embodiments, whether executing instructions stored on a non-transitory computer- readable storage medium or not, the processing circuitry can be configured to perform the described functionality. The benefits provided by such functionality are not limited to the processing circuitry alone or to other components of the computing device, but are enjoyed by the computing device as a whole, and/or by end users and a wireless network generally.

EMBODIMENTS

1. A method performed by a user equipment, UE, (1600) in a network, the method comprising: obtaining (501) a power allocation matrix derived based on a machine learning process, wherein each row of the power allocation matrix defines a power allocation vector specifying how much power is allocated to each beam of a number of access points, APs, each AP having a number of beam directions per AP; measuring (503) combined reference signals received by the UE for each power allocation combination transmitted according to the power allocation matrix; calculating (505) a sparse vector D using a compressed sensing algorithm based on a number of the combined reference signals measured; determining (507) best beam directions based on the sparse vector D; and signaling (509) the best beam directions towards a network node in the network.

2. The method of Embodiment 1, wherein measuring combined reference signals received by the UE comprises measuring a received composite signal strength for each power allocation combination.

3. The method of any of Embodiments 1-2, wherein calculating a sparse vector D using a compressed sensing algorithm comprises calculating a sparse vector D using one of a orthogonal matching pursuit, OMP, an iterative hard thresholding, IHT, and a basis pursuit, BP.

4. The method of any of Embodiments 1-3, wherein signaling the best beam directions towards the network node comprises signaling vector indexes of beam indexes that are above a threshold towards the network node.

5. The method of any of Embodiments 1-4, wherein calculating a sparse vector D using a compressed sensing algorithm based on a number of the combined reference signals measured comprises iteratively calculating the sparse vector D using the compressed sensing algorithm based on different numbers of the combined reference signals measured

6. The method of any of Embodiments 1-5, wherein determining the best beam directions comprises: determining which beam directions are above a threshold; and indicating that the beam directions that are above the threshold are the best beam directions.

7. The method of any of Embodiments 1-6, wherein calculating the sparse vector D using the compressed sensing algorithm comprises calculating the sparse vector D based on a machine learning process.

8. The method of any of Embodiments 1-7, wherein calculating the sparse vector D using the compressed sensing algorithm comprises calculating the sparse vector D based on dictionary learning.

9. A method performed by a network node (1700) in a network, the method comprising: transmitting (601) a plurality of reference signals on beams on a plurality of same timefrequency resources in all directions in accordance with power allocation vectors of a power allocation matrix; receiving (603), from a UE, composite signal strengths received for each power allocation vector transmitted; calculating (605) a sparse vector D using a compressed sensing algorithm based on a number of the composite signal strengths measured; and determining (607) best beams to use to transmit to the UE based on the sparse vector D.

10. The method of Embodiment 9, further comprising determining the power allocation matrix based on a machine learning process, wherein each row of the power allocation matrix defines a power allocation vector specifying how much power is allocated to each beam of a number of access points, APs, each AP having a number of beam directions per AP;

11. The method of Embodiment 10, wherein determining the power allocation matrix comprises determining the power allocation matrix based on dictionary learning.

12. The method of Embodiment 11, wherein determining the power allocation matrix based on dictionary learning comprises: collecting learning measurement samples where the sparse vectors D are directly measured; determining an optimal power allocation matrix such that the sparse vectors D can be reconstructed from a shortest possible measurement vector. 13. The method of Embodiment 11, wherein determining the power allocation matrix based on dictionary learning comprises determining the power allocation matrix using an autoencoder neural network.

14. The method of any of Embodiments 9-13, further comprising: projecting a set of sparse vector codes into a lower dimensional sub-space via one or more dimensionality reduction techniques; determining a power allocation sub-matrix in the lower dimensional sub-space; deriving the power allocation matrix by determining a pseudo-inverse of the power allocation sub-matrix.

15. The method of any of Embodiments 9-14, further comprising normalizing a power of beams corresponding to a same AP to AP total transmit power.

16. The method of any of Embodiments 9-15, wherein calculating a sparse vector D using a compressed sensing algorithm comprises calculating a sparse vector D using one of an orthogonal matching pursuit, OMP, an iterative hard thresholding, IHT, and a basis pursuit, BP.

17. The method of any of Embodiments 9-16, wherein calculating a sparse vector D using a compressed sensing algorithm based on a number of the combined reference signals measured comprises iteratively calculating the sparse vector D using the compressed sensing algorithm based on different numbers of the combined reference signals measured.

18. The method of any of Embodiments 9-17, wherein determining the best beam directions comprises: determining which beam directions are above a threshold; and indicating that the beam directions thar are above the threshold are the best beam directions.

19. A user equipment, UE, (1600) adapted to operate in a network, the UE comprising: a processor (1602); and memory (1610) coupled with the processing circuitry, wherein the memory includes instructions that when executed by the processing circuitry causes the UE to perform operations comprising: obtaining (501) a power allocation matrix derived based on a machine learning process, wherein each row of the power allocation matrix defines a power allocation vector specifying how much power is allocated to each beam of a number of access points, APs, each AP having a number of beam directions per AP; measuring (503) combined reference signals received by the UE for each power allocation combination transmitted according to the power allocation matrix; calculating (505) a sparse vector D using a compressed sensing algorithm based on a number of the combined reference signals measured; determining (507) best beam directions based on the sparse vector D; and signaling (509) the best beam directions towards a network node in the network.

20. The UE (1600) of Embodiment 19, wherein the memory includes further instructions that when executed by the processing circuity causes the UE (1600) to perform operations in accordance with Embodiments 2-8.

21. A user equipment, UE (1600) adapted to perform operations comprising: obtaining (501) a power allocation matrix derived based on a machine learning process, wherein each row of the power allocation matrix defines a power allocation vector specifying how much power is allocated to each beam of a number of access points, APs, each AP having a number of beam directions per AP; measuring (503) combined reference signals received by the UE for each power allocation combination transmitted according to the power allocation matrix; calculating (505) a sparse vector D using a compressed sensing algorithm based on a number of the combined reference signals measured; determining (507) best beam directions based on the sparse vector D; and signaling (509) the best beam directions towards a network node in the network.

22. The UE of Embodiment 21, wherein the UE (1600) is adapted to perform operations in accordance with Embodiments 2-8.

23. A computer program comprising program code to be executed by processing circuitry (1602) of a user equipment, UE (1600), whereby execution of the program code causes the UE (1600) to perform operations according to any of Embodiments 2-8.

24. A computer program product comprising a non-transitory storage medium including program code to be executed by processing circuitry (Q202) of a user equipment, UE (1600), whereby execution of the program code causes the UE (1600) to perform operations according to any of Embodiments 2-8. 25. A network node (1700) adapted to operate in a network, the UE comprising: a processor (1702); and memory (1704) coupled with the processing circuitry, wherein the memory includes instructions that when executed by the processing circuitry causes the UE to perform operations comprising: transmitting (601) a plurality of reference signals on the beams on a plurality of same time-frequency resource in all directions in accordance with power allocation vectors of a power allocation matrix; receiving (603), from a UE, composite signal strengths received for each power allocation vector transmitted; calculating (605) a sparse vector D using a compressed sensing algorithm based on a number of the composite signal strengths measured; and determining (607) best beams to use to transmit to the UE based on the sparse vector D.

26. The network node (1700) of Embodiment 25, wherein the memory includes further instructions that when executed by the processing circuity causes the UE (1600) to perform operations in accordance with Embodiments 10-18.

27. A network node (1700) adapted to perform operations comprising: transmitting (601) a plurality of reference signals on the beams on a plurality of same time-frequency resource in all directions in accordance with power allocation vectors of a power allocation matrix; receiving (603), from a UE, composite signal strengths received for each power allocation vector transmitted; calculating (605) a sparse vector D using a compressed sensing algorithm based on a number of the composite signal strengths measured; and determining (607) best beams to use to transmit to the UE based on the sparse vector D.

28. The network node (1700) of Embodiment 27, wherein the network node (1700) is adapted to perform operations in accordance with Embodiments 10-18.

29. A computer program comprising program code to be executed by processing circuitry (1702) of a network node (1700), whereby execution of the program code causes the network node (1700) to perform operations according to any of Embodiments 10-18. 30. A computer program product comprising a non-transitory storage medium including program code to be executed by processing circuitry (1702) of a network node (1700), whereby execution of the program code causes the UE network node (1700) to perform operations according to any of Embodiments 10-18.

Claims

CLAIMS What is claimed is:

2. The method of claim 1, wherein measuring combined reference signals received by the UE comprises measuring a received composite signal strength for each power allocation combination.

3. The method of any of claims 1-2, wherein calculating a sparse vector D using a compressed sensing algorithm comprises calculating a sparse vector D using one of a orthogonal matching pursue, OMP, an iterative hard thresholding, IHT, and a basis pursuit, BP.

4. The method of any of claims 1-3, wherein signaling the best beam directions towards the network node comprises signaling vector indexes of beam indexes that are above a threshold towards the network node.

5. The method of any of claims 1-4, wherein calculating a sparse vector D using a compressed sensing algorithm based on a number of the combined reference signals measured comprises iteratively calculating the sparse vector D using the compressed sensing algorithm based on different numbers of the combined reference signals measured.

6. The method of any of claims 1-5, wherein determining the best beam directions comprises: determining which beam directions are above a threshold; and indicating that the beam directions that are above the threshold are the best beam directions.

7. The method of any of claims 1-6, wherein calculating the sparse vector D using the compressed sensing algorithm comprises calculating the sparse vector D based on a machine learning process.

8. The method of any of claims 1-7, wherein calculating the sparse vector D using the compressed sensing algorithm comprises calculating the sparse vector D based on dictionary learning.

9. The method of any of claims 1-8, wherein obtaining the power allocation matrix derived based on the machine learning process comprises obtaining the power allocation matrix using an autoencoder architecture with a neural network-based sparse decoder.

10. A method performed by a network node (1700) in a network, the method comprising: transmitting (601) a plurality of reference signals on beams on a plurality of same timefrequency resources in all directions in accordance with power allocation vectors of a power allocation matrix; receiving (603), from a UE, composite signal strengths received for each power allocation vector transmitted; calculating (605) a sparse vector D using a compressed sensing algorithm based on a number of the composite signal strengths measured; and determining (607) best beams to use to transmit to the UE based on the sparse vector D.

11. The method of claim 10, further comprising determining the power allocation matrix based on a machine learning process, wherein each row of the power allocation matrix defines a power allocation vector specifying how much power is allocated to each beam of a number of access points, APs, each AP having a number of beam directions per AP.

12. The method of claim 11, wherein determining the power allocation matrix comprises determining the power allocation matrix based on dictionary learning.

13. The method of claim 12, wherein determining the power allocation matrix based on dictionary learning comprises: collecting learning measurement samples where the sparse vectors D are directly measured; determining an optimal power allocation matrix such that the sparse vectors D can be reconstructed from a shortest possible measurement vector.

14. The method of claim 12, wherein determining the power allocation matrix based on dictionary learning comprises determining the power allocation matrix using an autoencoder neural network.

15. The method of any of claims 10-14, further comprising: projecting a set of sparse vector codes into a lower dimensional sub-space via one or more dimensionality reduction techniques; determining a power allocation sub-matrix in the lower dimensional sub-space; deriving the power allocation matrix by determining a pseudo-inverse of the power allocation sub-matrix.

16. The method of any of claims 10-15, further comprising normalizing a power of beams corresponding to a same AP to AP total transmit power.

17. The method of any of claims 10-16, wherein calculating a sparse vector D using a compressed sensing algorithm comprises calculating a sparse vector D using one of an orthogonal matching pursue, OMP, an iterative hard thresholding, IHT, and a basis pursuit, BP.

18. The method of any of claims 10-17, wherein calculating a sparse vector D using a compressed sensing algorithm based on a number of the combined reference signals measured comprises iteratively calculating the sparse vector D using the compressed sensing algorithm based on different numbers of the combined reference signals measured.

19. The method of any of claims 10-18, wherein determining the best beam directions comprises: determining which beam directions are above a threshold; and indicating that the beam directions thar are above the threshold are the best beam directions.

20. A user equipment, UE, (1600) adapted to operate in a network, the UE comprising: a processor (1602); and memory (1610) coupled with the processing circuitry, wherein the memory includes instructions that when executed by the processing circuitry causes the UE to perform operations comprising: obtaining (501) a power allocation matrix derived based on a machine learning process, wherein each row of the power allocation matrix defines a power allocation vector specifying how much power is allocated to each beam of a number of access points, APs, each AP having a number of beam directions per AP; measuring (503) combined reference signals received by the UE for each power allocation combination transmitted according to the power allocation matrix; calculating (505) a sparse vector D using a compressed sensing algorithm based on a number of the combined reference signals measured; determining (507) best beam directions based on the sparse vector D; and signaling (509) the best beam directions towards a network node in the network.

21. The UE (1600) of claim 20, wherein the memory includes further instructions that when executed by the processing circuity causes the UE (1600) to perform operations in accordance with claims 2-9.

22. A user equipment, UE (1600) adapted to perform operations comprising: obtaining (501) a power allocation matrix derived based on a machine learning process, wherein each row of the power allocation matrix defines a power allocation vector specifying how much power is allocated to each beam of a number of access points, APs, each AP having a number of beam directions per AP; measuring (503) combined reference signals received by the UE for each power allocation combination transmitted according to the power allocation matrix; calculating (505) a sparse vector D using a compressed sensing algorithm based on a number of the combined reference signals measured; determining (507) best beam directions based on the sparse vector D; and signaling (509) the best beam directions towards a network node in the network.

23. The UE of claim 22, wherein the UE (1600) is adapted to perform operations in accordance with claims 2-9.

24. A computer program comprising program code to be executed by processing circuitry (1602) of a user equipment, UE (1600), whereby execution of the program code causes the UE (1600) to perform operations according to any of claims 1-9.

25. A computer program product comprising a non-transitory storage medium including program code to be executed by processing circuitry (Q202) of a user equipment, UE (1600), whereby execution of the program code causes the UE (1600) to perform operations according to any of claims 1-9.

26. A network node (1700) adapted to operate in a network, the UE comprising: a processor (1702); and memory (1704) coupled with the processing circuitry, wherein the memory includes instructions that when executed by the processing circuitry causes the UE to perform operations comprising: transmitting (601) a plurality of reference signals on the beams on a plurality of same time-frequency resource in all directions in accordance with power allocation vectors of a power allocation matrix; receiving (603), from a UE, composite signal strengths received for each power allocation vector transmitted; calculating (605) a sparse vector D using a compressed sensing algorithm based on a number of the composite signal strengths measured; and determining (607) best beams to use to transmit to the UE based on the sparse vector D.

27. The network node (1700) of claim 26, wherein the memory includes further instructions that when executed by the processing circuity causes the UE (1600) to perform operations in accordance with claims 11-19.

28. A network node (1700) adapted to perform operations comprising: transmitting (601) a plurality of reference signals on the beams on a plurality of same time-frequency resource in all directions in accordance with power allocation vectors of a power allocation matrix; receiving (603), from a UE, composite signal strengths received for each power allocation vector transmitted; calculating (605) a sparse vector D using a compressed sensing algorithm based on a number of the composite signal strengths measured; and determining (607) best beams to use to transmit to the UE based on the sparse vector D.

29. The network node (1700) of claim 27, wherein the network node (1700) is adapted to perform operations in accordance with claims 11-19.

30. A computer program comprising program code to be executed by processing circuitry (1702) of a network node (1700), whereby execution of the program code causes the network node (1700) to perform operations according to any of claims 10-19.

31. A computer program product comprising a non-transitory storage medium including program code to be executed by processing circuitry (1702) of a network node (1700), whereby execution of the program code causes the UE network node (1700) to perform operations according to any of claims 10-19.